Personalized Clothing-Recommendation System Based on a Modified Bayesian Network
Yu-chu Lin, Yuusuke Kawakita, Etsuko Suzuki, Haruhisa Ichikawa · 2012
This paper presents a clothing-recommendation system that suggests personal combinations from a user's wardrobe. Online shopping websites use recommendation systems to suggest items that users might be interested in. Such systems make recommendations to a user on the basis of other users' behavior, under the assumption that all users behave similarly; personal preferences are not captured. However, a clothing-recommendation system should make recommendations on selections of personal items based on personal preferences rather than other users' behavior, since it is rare to find other users that own the same articles of clothing as the target user. The proposed system makes recommendations that are particularly suitable to a user based on the user's personal preference, history of clothing items and the user's evaluations of previous system recommendations. The experimental results reveal that the system can in most cases recommend more suitable combinations of clothing items than can existing systems under the same conditions.